ABSTRACT
Supposing that an appropriate epistemic end can be identified on the basis of such epistemic values, the next question to face is: "How, in relation to such an end, can standards for epistemic behavior be identified?" Since our human condition is, as Quine put it, a Humean condition,34 we cannot expect to identify standards that, if conscientiously adhered to, will ensure the attainment of such an end. We must allow that the relation between standards and end will be weaker than this. I suggest that, if the end is to obtain a maximally comprehensive yet maximally simple conceptual scheme that can reasonably be hoped to be minimally erroneous, two sorts of epistemic standards are appropriate. The first prescribes policies that actively promote the end. To obtain the intended comprehensiveness in one's conceptual scheme, for example, one should seek "causes" or explanations for observed phenomena; and to obtain the intended simplicity and systematic connection in the scheme, one should minimize redundancy, attempt to subsume disparate phenomena under common principles, eliminate nomological danglers, and so forth. Conceptual activity will be epistemically permissible only when it accords with such policies. Standards of the second sort promote the end indirectly - by prohibiting things that will frustrate it. The key policy here is consistency, both deductive and inductive. As I have explained, the basis for the latter is the probability calculus.
